Gossamery Arguments for Transparency and Our Reply

Recently, we’ve seen many op-ed essays calling for more transparency in financial statements, particularly with respect to mortgage-related securities.  Many of these essays have been written by famous and esteemed individuals or their staffs.

In our own idiosyncratic, round-about way, we’ll explain the empty silliness of such arguments, and we begin by criticizing the notion that “more is always better.”

Too Much Information: Unfortunately, we’ve not read a single essay that contained an intelligent, concrete argument for why more transparency is better than less–as if transparency, in and of itself, is a good (or is inherently good).

More precisely, in all of these articles, the value of transparency is assumed, and the assumption seems to be implicit and subconscious (unconscious?) rather than something arrived at via serious deliberation.  (Hint: we can’t recall any of these essays that bother to define transparency.  Presumably, it is like pornography: you know it when you see it.)

In that half-assed way, these recent prompts for more transparency have much in common with the slightly older admonitions to eliminate mark-to-market accounting.¹

In their theories, many economists–including, yours truly–have shown that more transparency, which often means more precise information, is not always better than less; in fact, it can make things strictly worse.  Such seemingly pathological results are actually rather common in a variety of social settings, including some markets, and arise for a number of reasons, including risk-sharing and incentives, where more information can affect an agent’s behavior and actions or efforts thereby reducing social welfare and/or exacerbating incentive problems.

For example (and this is a gross generalization of the results without specifying any of the necessary assumptions) in Kanodia, Singh and Spero (JAR, 2005), we show that it is better to keep two unknown variables as unknowns rather than know only one with perfect precision.  Think of it in the following way: suppose there are two random variables–one that is somewhat in the person’s control and the other, which is not.

If the one under his influence is known perfectly, he’ll overemphasize it.  If the other one is known perfectly, then he’ll rightfully conclude that the noisy signal of his effort will be overlooked in favor of the other variable so he’ll do little.  The former creates over-exertion and the latter creates under-exertion and both are socially damaging; thus, one can find a happy medium in less extreme cases where neither variable is known with total precision.  (It should remind one of Goldilocks.)

Now, let’s be very clear that one need not be an economist to know that more information or transparency is not always better.  For example, how does the reader answer questions from a spouse, relative, or friend when asked something like, “Do you like my new haircut?” or “Does this dress make me look fat?”

In addition, there are other cases where another party reveals personal details with too much precision.  In fact, we as a society have the colloquialism, “Too much information!” for just such cases where you’ll never again look at the revealer in the same manner and subsequently ruefully wonder, “why did they have to tell me that?”

Details Are Not Information: this is a particularly apt time to repeat our admonition that details are not information.  Back in April, we posted a long essay on the difference between details and information or useful facts.  (Useful facts are ones that might cause a decision to change as the fact is realized.)  Our point in that essay was to distinguish between keeping track of a lot of necessary data–as in data processing–and the quite different task of providing useful information to decision-makers.  If one leaves systems design to systems people, one will likely get the former and not much of the latter.  Moreover, if the decision-maker can’t design the system–not the programming–then his or her competence at decision-making should be justifiably questioned.

The same distinction between details and information holds true with financial assets, too.  More transparency can mean an inundation of book-keeping and account details, which may provide no information or which may require expert judgment to (sift through to) become information.  In either case, the recipient of the data dump may not “see the forest for the trees.”²  So, one may have all the facts, but no ability to organize them–much like a child writing a term paper.

And, that, of course, illustrates the silliness of calling for more transparency for mortgage-related securities.  The bigger problem is that with every datum about every mortgage in a pool, there is still no easy way to value them.

The issue isn’t the details, it is how to combine current and past details to determine value and risk in the future, and it is very likely a perfect method is unknowable.  So…

Value Matters, BUT There’s No Transparent Way to Find It: let’s illustrate the notion in to a fairly high level of detail (for a blog post).  We’ll ignore the “waterfall” aspect of real mortgage-backed securities and CDOs where different classes of security holders have different priority claims on the cash flows because those claims are not the confounding factors–the interelationships of the mortgages are.

So, imagine a pool of T thousand mortgages going down the first column of a spreadsheet. Further, suppose that the next 360 columns represent months, m, so, the row t and column m intersection is the amount of cash received from borrower t in month m.  Now that cell will actually be a function of any number of factors, including interest rates which affect whether the mortgage is repaid early; the person’s wealth and income which determine whether the borrower declares bankruptcy, the relationship between the value of the collateral and the loan balance, etc.  We could go on and on, but the point is that each cell could take any number of values depending upon many different factors.

One page of the spreadsheet would then represent one entire scenario of how cash is received from all T thousand mortgages over the next thirty years.

At issue for valuation (and risk modeling) is how to combine outcomes across all mortgages.  The cells are clearly related within a row, i.e., a borrower’s status in one month will affect cash flows in later months.

But, cash flows are also related within columns–phenomena, like a hurricane, may contemporaneously affect more than one borrower–and across columns, too.  For example, someone’s default in month m may make another’s default in month m + n more likely.  So, the bigger issue is: how does one relate borrowers across time and space to arrive at a distribution of cash flows.  (Note: we mean “space” literally because community and regional effects matter–the inter-row action, sometimes.)

One could generate any number of scenarios or pages, but, of course, the issue for valuation (and risk) are which combinations in the numerous T x 360 grid are more (or less) likely (and how wide is the range of possible outcomes)?

In other words, the problem lays with determining the joint distributions across borrowers and time.  As we see it, there is no correct method, but there is an infinity of incorrect methods, especially ones that rely only on historical relationships, particularly very short histories.

Those incorrect methods include many that were implemented in recent years.  As we see it, many of those methods were implemented because they were solvable, not because they were accurate.  Unfortunately, those weaknesses (inaccuracies) were obscured by the relative calmness of the markets, including the near-Ponzi-like schemes of different banks buying the securities to re-securitize them yet another time.

So, we ask those writers urging more transparency: exactly how would it help us find a price in the above example?  Our illustration highlights the reason why there is a lack of buyers.  There are data aplenty.  What is lacking is a quantifiable notion of the future.

That, dear reader, is why we developed and wrote about an alternative solution to TARP.  One that involved the use of investment tax credits or cash-basis accounting (to permit the immediate expense of the purchase price) to subsidize and cushion the risk of purchasing these conglomerations of cash flows.  It would provide private buyers with an immediate benefit of 30% – 40% of the purchase price, which seems large enough to permit room for error.

As always, we encourage visitors to read our essay, Uncertainty Management, which discusses the notions of measurability (quantifiability) and immeasurability by distinguishing between the broader idea of uncertainty and the narrower idea of risk.  In that regard, the number and cost of mis-specification errors related to our ongoing crisis may be the greatest in any period in history.

We’ll probably edit this again in the near future.

Footnotes: